Our friends at PC INpact have contacted AMD about FurMark detection in the latest Catalyst drivers.
AMD has replied that FurMark’s slowdown is really intentional because softwares such as FurMark damage graphics cards and do not represent real world applications.
We all know this tweak in Catalyst was deliberate since it has been first discovered by Expreview last year. But now it’s the first time it’s official from AMD. It’s fun, the first protection has been added in Catalyst 8.8 and the second update of this protection has been added in Catalyst 9.8. Maybe FurMark will be totally banned in Catalyst 10.8 🙂
I hope this second update does not hide severe problems with the upcoming Radeon HD 5000 series…
Can you imagine Intel to slowdown the CPU because Stress Prime or 3DMark Vantage CPU push the processor too far?
I coded the original FurMark algorithm in a couple of hours (and I’m not a hardcore graphics programming guru) without the goal to burn my graphics card. The overheat produced by the fur rendering is a simple consequence of the multipass nature of the algorithm (this discovery led to the first version of FurMark in August 2007). So if a rendering algorithm points out some hardware problems, the easy solution seems to get ride of it.
There are many multipass algorithms and I’m sure we’ll see soon or later some that will be able to burn the GPU in the same way as FurMark (currently OCCT GPU does the job very well too!). By the way I wonder if OCCT GPU is detected as a power virus by Catalyst? And what will occur if, for example, Carmack, in his next 3D engine, codes a cool effect that burns the GPU?
I tested for hours FurMark with my HIS’s Radeon HD 4850 without any overheat problem (with default clocks and with overclocking). Ok GPU temperature exceeded 90°C but that’s all. So what? Is HIS the unique ATI’s partner to release good quality graphics cards?
A simple solution would have been AMD to give me the list of problematic video cards in order I add an official protection in FurMark when these cards are used.
Currently I don’t have time to find a workaround, but I’ll do it asap!